phoenix-oss/llama-stack
1
0
Fork 0
forked from phoenix-oss/llama-stack-mirror
Code Pull requests Releases Packages 2 Activity Actions
1819 commits 21 branches 64 tags 62 MiB
Commit graph

3 commits

Author SHA1 Message Date
Jash Gulabrai
8713d67ce3
fix: Correctly parse algorithm_config when launching NVIDIA customization job; fix internal request handler (#2025) 
# What does this PR do?
This addresses 2 bugs I ran into when launching a fine-tuning job with
the NVIDIA Adapter:
1. Session handling in `_make_request` helper function returns an error.
```
INFO:     127.0.0.1:55831 - "POST /v1/post-training/supervised-fine-tune HTTP/1.1" 500 Internal Server Error
16:11:45.643 [END] /v1/post-training/supervised-fine-tune [StatusCode.OK] (270.44ms)
 16:11:45.643 [ERROR] Error executing endpoint route='/v1/post-training/supervised-fine-tune' method='post'
Traceback (most recent call last):
  File "/Users/jgulabrai/Projects/forks/llama-stack/llama_stack/distribution/server/server.py", line 201, in endpoint
    return await maybe_await(value)
  File "/Users/jgulabrai/Projects/forks/llama-stack/llama_stack/distribution/server/server.py", line 161, in maybe_await
    return await value
  File "/Users/jgulabrai/Projects/forks/llama-stack/llama_stack/providers/remote/post_training/nvidia/post_training.py", line 408, in supervised_fine_tune
    response = await self._make_request(
  File "/Users/jgulabrai/Projects/forks/llama-stack/llama_stack/providers/remote/post_training/nvidia/post_training.py", line 98, in _make_request
    async with self.session.request(method, url, params=params, json=json, **kwargs) as response:
  File "/Users/jgulabrai/Projects/forks/llama-stack/.venv/lib/python3.10/site-packages/aiohttp/client.py", line 1425, in __aenter__
    self._resp: _RetType = await self._coro
  File "/Users/jgulabrai/Projects/forks/llama-stack/.venv/lib/python3.10/site-packages/aiohttp/client.py", line 579, in _request
    handle = tm.start()
  File "/Users/jgulabrai/Projects/forks/llama-stack/.venv/lib/python3.10/site-packages/aiohttp/helpers.py", line 587, in start
    return self._loop.call_at(when, self.__call__)
  File "/Library/Frameworks/Python.framework/Versions/3.10/lib/python3.10/asyncio/base_events.py", line 724, in call_at
    self._check_closed()
  File "/Library/Frameworks/Python.framework/Versions/3.10/lib/python3.10/asyncio/base_events.py", line 510, in _check_closed
    raise RuntimeError('Event loop is closed')
RuntimeError: Event loop is closed
```
Note: This only occurred when initializing the client like so:
```
client = LlamaStackClient(
    base_url="http://0.0.0.0:8321"
)
response = client.post_training.supervised_fine_tune(...) # Returns error
```
I didn't run into this issue when using the library client:
```
client =  LlamaStackAsLibraryClient("nvidia")
client.initialize()
response = client.post_training.supervised_fine_tune(...) # Works fine
```

2. The `algorithm_config` param in `supervised_fine_tune` is parsed as a
`dict` when run from unit tests, but a Pydantic model when invoked using
the Llama Stack client. So, the call fails outside of unit tests:
```
INFO:     127.0.0.1:54024 - "POST /v1/post-training/supervised-fine-tune HTTP/1.1" 500 Internal Server Error
21:14:02.315 [END] /v1/post-training/supervised-fine-tune [StatusCode.OK] (71.18ms)
 21:14:02.314 [ERROR] Error executing endpoint route='/v1/post-training/supervised-fine-tune' method='post'
Traceback (most recent call last):
  File "/Users/jgulabrai/Projects/forks/llama-stack/llama_stack/distribution/server/server.py", line 205, in endpoint
    return await maybe_await(value)
  File "/Users/jgulabrai/Projects/forks/llama-stack/llama_stack/distribution/server/server.py", line 164, in maybe_await
    return await value
  File "/Users/jgulabrai/Projects/forks/llama-stack/llama_stack/providers/remote/post_training/nvidia/post_training.py", line 407, in supervised_fine_tune
    "adapter_dim": algorithm_config.get("adapter_dim"),
  File "/Users/jgulabrai/Projects/forks/llama-stack/.venv/lib/python3.10/site-packages/pydantic/main.py", line 891, in __getattr__
    raise AttributeError(f'{type(self).__name__!r} object has no attribute {item!r}')
AttributeError: 'LoraFinetuningConfig' object has no attribute 'get'
```
The code assumes `algorithm_config` should be `dict`, so I just handle
both cases.

[//]: # (If resolving an issue, uncomment and update the line below)
[//]: # (Closes #[issue-number])

## Test Plan
1. I ran a local Llama Stack server with the necessary env vars:
```
lama stack run llama_stack/templates/nvidia/run.yaml --port 8321 --env ...
```
And invoked `supervised_fine_tune` to confirm neither of the errors
above occur.
```
client = LlamaStackClient(
    base_url="http://0.0.0.0:8321"
)
response = client.post_training.supervised_fine_tune(...)
```
2. I confirmed the unit tests still pass: `./scripts/unit-tests.sh
tests/unit/providers/nvidia/test_supervised_fine_tuning.py`

[//]: # (## Documentation)

---------

Co-authored-by: Jash Gulabrai <jgulabrai@nvidia.com>
 2025-04-25 13:21:50 -07:00
Ihar Hrachyshka
193e531216
chore: re-enable isort enforcement (#1802) 
# What does this PR do?

Re-enable isort enforcement.

It was disabled in 1a73f8305b, probably by
mistake.

Signed-off-by: Ihar Hrachyshka <ihar.hrachyshka@gmail.com>
 2025-03-26 15:22:17 -07:00
Rashmi Pawar
1a73f8305b
feat: Add nemo customizer (#1448) 
# What does this PR do?

This PR adds support for NVIDIA's NeMo Customizer API to the Llama Stack
post-training module. The integration enables users to fine-tune models
using NVIDIA's cloud-based customization service through a consistent
Llama Stack interface.


[//]: # (If resolving an issue, uncomment and update the line below)
[//]: # (Closes #[issue-number])

## Test Plan
[Describe the tests you ran to verify your changes with result
summaries. *Provide clear instructions so the plan can be easily
re-executed.*]
Yet to be done

Things pending under this PR:

- [x] Integration of fine-tuned model(new checkpoint) for inference with
nvidia llm distribution
- [x] distribution integration of API
- [x] Add test cases for customizer(In Progress)
- [x] Documentation

```

LLAMA_STACK_BASE_URL=http://localhost:5002 pytest -v tests/client-sdk/post_training/test_supervised_fine_tuning.py 

============================================================================================================================================================================ test session starts =============================================================================================================================================================================
platform linux -- Python 3.10.0, pytest-8.3.4, pluggy-1.5.0 -- /home/ubuntu/llama-stack/.venv/bin/python
cachedir: .pytest_cache
metadata: {'Python': '3.10.0', 'Platform': 'Linux-6.8.0-1021-gcp-x86_64-with-glibc2.35', 'Packages': {'pytest': '8.3.4', 'pluggy': '1.5.0'}, 'Plugins': {'nbval': '0.11.0', 'metadata': '3.1.1', 'anyio': '4.8.0', 'html': '4.1.1', 'asyncio': '0.25.3'}}
rootdir: /home/ubuntu/llama-stack
configfile: pyproject.toml
plugins: nbval-0.11.0, metadata-3.1.1, anyio-4.8.0, html-4.1.1, asyncio-0.25.3
asyncio: mode=strict, asyncio_default_fixture_loop_scope=None
collected 2 items                                                                                                                                                                                                                                                                                                                                                            

tests/client-sdk/post_training/test_supervised_fine_tuning.py::test_post_training_provider_registration[txt=8B] PASSED                                                                                                                                                                                                                                                 [ 50%]
tests/client-sdk/post_training/test_supervised_fine_tuning.py::test_list_training_jobs[txt=8B] PASSED                                                                                                                                                                                                                                                                  [100%]

======================================================================================================================================================================== 2 passed, 1 warning in 0.10s ========================================================================================================================================================================
```
cc: @mattf @dglogo @sumitb

---------

Co-authored-by: Ubuntu <ubuntu@llama-stack-customizer-dev-inst-2tx95fyisatvlic4we8hidx5tfj.us-central1-a.c.brevdevprod.internal>
 2025-03-25 11:01:10 -07:00